Peter and all, Good points, and You got 4 1/2 out of five correct, Not bad at all.
#1) YES, The easy way I used to make the controlled phase error steps was by entering a new number into the cable delay, which is limited to under 50ns change. #2) Yes, this is the self reported Tbolt data plotted by LH. #3) Yes, All the Thunderbolt can do is report the differences between itself and the (short term) noisy GPS signal. #4) Yes, what is really desired is a comparison to a stable external standard. #5) Yes, Yes&NO, No, ... Such a comparison would give a rather different plot. 5a) Very much Yes, when plotting unfiltered display data, (The New LH can filter the plots, should use 100 + sec to minimize the GPS noise) 5b) somewhat Yes looking at times under 500 sec or so. and the important thing, and the only thing I was trying to show is: 5c) NO for data longer than 1000 sec, where the GPS is going to be pretty much the most accurate thing that most have. So with the plots I made, showing the Basic long term TC time and shape, they are as close as anything you are going to get using an external ref. On the other hand, The high freq wiggles in the 0 to 500 sec range, are not real. They are the noisy GPS and not the OSC. If you want to run the same test at faster TC or with more accuracy, JUST need to increase the Phase error by about times ten, so that the GPS nose is NOT a limiting factor. Will give the same basic shape, just less GPS noise wiggles. The basic shape of what I'm showing on the plots is valid, More so in fact than you would get with most external References. Concerning Toms ADEV numbers. Two different uses, His Plots show what the noise across the whole time span is, at the time the data was taken, These plots show Why, and by changing the filter setting show the relative noise at different time spans with different setting, with Post processing. ws *************** Thank you for your explanation. Your Lady Heather plot prompts a couple of questions though: 1) Presumably you created the 40ns phase step by changing the output timing offset? 2) Presumably, again, the plotted data is from the self reported data from the Thunderbolt? I have been running a series of tests myself this year, but on looking at the results, it became clear - and I should have twigged this earlier - that all the Thunderbolt can do is report the differences between itself and the noisy GPS signal. Whereas what I really wanted was a comparison to a stable external standard. Sadly I don't have TVB's Maser or 5071s, but I suspect that such a comparison would give a rather different plot. Tom has talked about these factors on his page at http://www.LeapSecond.com/pages/tbolt-tc/ just showing MDEV plots. Tom: would you have the time to replicate Warren's experiment, but show the phase and frequency plots with respect to your Maser or 5071? As an aside, at the beginning of the year when the TAPR offer was on, it was suggested on here that the time constant be set to that where the ADEV plot reaches a minimum whilst on holdover. A couple of days after turning on, mine was about 700 seconds. Now, nine months continuous operation later, and with the Thunderbolt in the recommended cardboard box to protect from drafts, the plot now reaches a minimum at about 2500 seconds! Peter _______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.